Chuck Carson <ccarson_No.Spammo_at_echeeba.com> schreef in berichtnieuws
3EBD7028.1000002_at_echeeba.com...
|
| So if I shut the database down and then startup mount, do I then drop
| the redo file? If so, doesn't this file still contain tranactions?
|
| -Chuck
There is no need to shutdown the database. You can drop redologs in an open
and active db (but when redolog switches occur very frequent it can be
difficult to catch the proper moment).
Check v$log.status. You can't drop the current redolog. You can only drop
stale or inactive redologs.
And when the database runs in archive log mode v$log.archived must be YES
too.
When a log switch occurs a checkpoint begins (status=active) (and archiving
if turned on). When that's finished, the redo is not needed anymore and can
be dropped or in a normal situation re-used.
Oracle will give appropriate errors if dropping is not possible.
A minimum of 2 redos is required. When redolog fill at a high rate and there
are only 2 redo groups the db might need to wait on the checkpoint
completion (as Sybrand already stated). You better create 4 redolog groups
and create redolog files 1a and 3a on disk1, 1b and 3b on disk2, 2a and 4a
on disk3 and 2b and 4b on disk4. Write activity is than on 2 disks only
(i.e. 1a and 1b) and mayby read for archiving is on the other 2 disks (4a
and 4b). Apparently writing redo is so heavy on your system that you want to
dedicate IO bandwith for it and that's why 2 redo groups is very likely not
the best choice.
Received on Sat May 10 2003 - 17:22:39 CDT
